Most of what this page is about, is how to make sure you use the right tester, in the right way. Only like this, such results are meaningful to discuss about. To make this page not too difficult, anything that supplies details is presented with a link, you are much advised to read those links when you are interested.
What tester to use
Generally, tube testers are divided in parametric testers, quality testers and curve tracers. The greater part of historical testers are quality testers, whereas such results are difficult to reproduce on two random testers of the same model. Besides these are all over 60 years old, and normally uncalibrated ever since 1965. The best tester to use is a parametric tester, and we supply such test results with our tubes.Please read following points, to make sure you test the same way as we do, so test results can be compared best.
Cold anode testing with curve tracers.
This is generally a problem, because such testers fail the capability to heat up the anode at it's required working temperature. No tube, new made or NOS, is specified to work well with cold anode. This is very tedious to explain, because the curve tracer manufacturers spread false information about this, or ignore the problem. But it DOES matter. Like when we test 300B-XLS, we do so in the factory at 495, 100mA, and we read the result after thermal balance, which needs 10 minutes.
A digital curve tracer uses only very short pulses of 1/1000 of a second, which generates no heat in the anode, and the only heat produced is by the cathode heater. So any thermal balance in the tube will not occur. Moreover some errors only occur under great heat, like anode current drift, or grid emission. It should be clear, a "passed" grid emission test at cold anode has nothing to say, because this occurs only above 80% of maximum anode heat. Yet this is the most occurring error with NOS tubes like RCA 45, and also RCA 2A3 may have it. This comes close to fooling the user.
So even when tube curves may look very impressive, myself I can not see anything specific from those, unless the tube has is defective anyway. This would become visible from the curves running into saturation. However in such a case, the tube bias would be wrong, and distortion is terrible. To call such a tube bad, we need no curve tracer for that. Yet for early diagnoses, tube curves unveil no useful information of any kind. There is no funny shape, or anything else visible. Long before the tube curves become optically bad, the dynamic parameters of the tube begin to change, but unfortunately for curve tracer users, these parameters can not be judged well with cold anode..
All together, we must take into account, the lack of heat will result in an error in the range of 5...20%, depending on the tube type and use condition. Used tubes will have a larger error than new tubes. Yet, being aware of the shortcomings, curve tracers may be used for what they are worth.
Gm (transconductance) must be measured in auto bias.
For reasons of simplicity, most tube testers and software, use fixed bias voltage to test the tubes, however due to natural grid voltage tolerance, which has nothing do do with emission, this will result in random plate current still. A compare with average plate current (so the data sheet value) will only tell how much deviation there is, and not tell much about the condition of the tube. To avoid this problem, we test the tubes in auto bias, meaning the plate current is always the same. Like this, we achieve the same temperature of the tube always. The test results will be: the grid voltage needed for this, and transconductance resulting. So these two numbers (Ug, and Gm) are the test result, and plate current is always the same. Ug by itself represents the factory tolerance on this, which number is not very meaningful to judge by itself. However a CHANGE of Ug is very important to judge wear out of older tubes. For this, you just set the grid voltage to the same value as when the tube was new, and the loss of plate current will immediately tell the tube condition. It should be obvious, the original, individual Ug voltage of the tube must be used for this, and not the data sheet value, which is only an average number.
However to test transconductance, you need to set the plate current back to the value as written on the box. Reasons are explained in the following text:
Transconductance depends very much on the plate current and also anode heat. For that reason, if a transconductance measurement is done, with the intention to compare it with something, you must always use the same anode heat, and the same plate current. So a transconductance measurement under some 'random' conditions can not be compared with the data sheet value, and also not with the individual value of the tubes. In case, you want to compare it with the data sheet value, you need to adjust the tube for the same anode heat, and same anode current as in the data sheet. This is done by adjusting Ua, and Ug until these conditions are achieved. Yet a compare with the data sheet would in the end only show the deviation of the average value, which may be -30% to +40% with factory new tubes, depending on the type and the brand, and in the end this tells not much about in individual tube itself. Much more interesting, is to compare a random EML tube Gm value, with the value it had when it was new. This will directly tell the use condition. For that reason we write the test conditions on the box, which means: Anode Voltage and Anode Current to be set. The setting of Anode Current must be done by changing Ug. After waiting minimum 5 Minutes for thermal balance, Transconductance (Gm) can be tested. This is how we do it at the EML factory, and only like this you can compare tube with out test data.
What is good manufacturer data?
At we take pride in presenting correct tube data, which is written on each tube box in detail.
What is a recommended tube tester?
Why we use the AT1000: The only reason why we use the Amplitrex AT1000 is because it can be bought new, and it can just do what we need. There are many AT1000 around, and they work all the same, because the software was never revised or debugged ever since 1995. You do need to be well aware of the limitations this brings. The default mode is useless for our purpose, and in general too, but we will explain to how to tester can be used in another mode. Test results can be good, if you define clearly what the tester is supposed to do.